Nutrients and Salinity Influence Prymnesium parvum (UTEX LB 2797) Elicited Sublethal Toxicity in Pimephales promelas and Danio rerio

Abstract

The magnitude, frequency, and duration of harmful algal blooms (HABs) are increasing worldwide due to climate change and anthropogenic activities. Prymnesium parvum is a euryhaline and eurythermal HAB forming species that has expanded throughout North America, resulting in massive fish kills. Previous aquatic ecology and toxicology efforts supported an understanding of conditions resulting in P. parvum HABs and fish kills; however, the primary endpoint selected for these studies was acute mortality. Whether adverse sublethal responses to P. parvum occur in fish are largely unknown. To begin to address this question, molecular and biochemical oxidative stress (OS) biomarker responses and photomotor behavioral alterations were investigated in two common fish models, the fathead minnow (Pimephales promelas) and zebrafish (Danio rerio). Varying nutrient and salinity conditions influenced P. parvum related OS biomarkers and fish behavioral responses in zebrafish and fathead minnow models, which were heightened by nonoptimal conditions for P. parvum growth. Such sublethal observations present important considerations for future aquatic assessments and management of P. parvum HABs.

1. Introduction

The frequency, duration, and intensity of harmful algal blooms (HABs) are increasing at a global scale primarily due to eutrophication, urbanization, salinization, climate change, altered hydrology, and storm water and agricultural runoff (Hallegraeff, 1993; Granéli et al., 2008; Paerl and Huisman, 2009; Roelke et al., 2012). HABs often result from multiple interlinking factors (both anthropogenic and natural) typically specific to each algal species, which presents challenges when assessing and managing water quality threats from HABs (Brooks et al., 2016, 2017). Environmental impacts include reduced dissolved oxygen concentrations, altered community structure, and impairments to recreational, fisheries and source water uses, particularly when nuisance taste and odor compounds and toxins are produced (Paerl et al., 2001). Adverse outcomes resulting from HABs include human health illnesses and ecological damage leading to substantial threats to coastal and inland waters (Brooks et al., 2016, 2017; Grattan et al., 2016).

Prymnesium parvum is a globally relevant mixotrophic and invasive species capable of forming HABs leading to massive fish kills that result in large economic losses (Southard et al., 2010; Brooks et al., 2011; Roelke et al., 2016). P. parvum typically blooms in marine and brackish waters, though blooms have increasingly occurred in low saline inland lakes and reservoirs (0.5–5 salinity) (Roelke et al., 2016). In the United States (US), P. parvum was first observed in the Pecos River, Texas in 1985. Since this discovery, recurrent blooms of P. parvum have followed in Texas and other areas of the U.S. across latitudes and various climates (Brooks et al. 2011; Roelke et al., 2016). For example, P. parvum blooms have been documented as far north as Pennsylvania (Roelke et al. 2016). Successful invasion of P. parvum is attributed to toxin(s) production, which functions allelopathically to outcompete other algal species, an antipredatory mechanism to deter grazers, and/or as an aid for mixotrophic nutrition (Tillmann, 1998; Granéli and Johansson, 2003; Tillmann, 2003; Carvalho and Granéli, 2010; Brooks et al., 2011; Driscoll et al., 2013). In addition, P. parvum is euryhaline and eurythermal (e.g., salinities 10 times lower and 10 °C below optimal conditions); these suboptimal conditions appear to influence toxin(s) production (Baker et al., 2007, 2009).

Multiple toxins have been identified as prymnesins, however, identification and quantification of the full suite of toxins responsible for the blooms have not yet been identified but it remains an active area of research (Igarashi et al. 1996, 1999; Schug et al., 2010; Henrikson et al., 2010; Manning and La Claire, 2010; Bertin et al., 2012a, 2012b; Blossom et al., 2014a; Blossom et al., 2014b; Rasmussen et al. 2016; Hems et al., 2018; Binzer et al. 2019; Svenssen et al., 2019). Monitoring of P. parvum HABs has thus been a challenge due to lack of sufficient characterization of the toxins produced and lack of their corresponding analytical standards for robust quantitation. Lack of standards has further limited experimental environmental fate and toxicology research (Brooks et al., 2011). Subsequently, standardized aquatic toxicity studies are increasingly employed to determine the occurrence of bioavailable toxins (Brooks et al. 2010). Adding to the challenge of understanding and comparing P. parvum HAB toxicity is a lack in published standard culture procedures, leading to inconsistencies with culture conditions and use of different toxicological study methods (Brooks et al., 2010). This represents an important consideration because non-optimal environmental conditions for growth differentially influence ambient acute toxicity to fish and other aquatic organisms (Baker et al., 2007, 2009; Brooks et al. 2010; Valenti et al., 2010; Roelke et al., 2007, 2015).

In addition to acute mortality responses, sublethal endpoints associated with adverse outcomes in aquatic life are critical for robust environmental risk assessment and management (Bradbury et al. 2004; Ankley et al., 2010). For example, preliminary results revealed sublethal responses (oxidative stress (OS) biomarkers and fish behavior alterations) following exposure to P. parvum, grown under suboptimal growth conditions (low salinity), at cell densities (1,000–15,000 cells/mL) below those inducing acute mortality, which is currently the primary endpoint used to determine ambient toxicity for this common HAB causing species (Brooks et al., 2010). Traditional morphometric sublethal endpoints (e.g., growth, reproduction) are important indicators for adverse outcomes linked to environmental protection goals; however, these responses are relatively nonspecific regarding molecular and cellular pathways by which toxicity is elicited. Adverse outcome pathways (AOP) provide a conceptual framework linking molecular initiating events cascading across multiple levels of biological organization leading to adverse outcomes at the individual and population level (Ankley et al., 2010). An AOP was recently developed for P. parvum after oxygen consumption and respiration were affected in rainbow trout; however, the molecular initiating event(s) leading to osmotic gill damage and population level impacts remain unknown (Svendsen et al., 2018). Endpoints including fish behavioral alterations and biochemical and molecular markers have been examined for multiple environmental stressors, including algal toxins (Drummond and Russom, 1990; Valavanidis et al., 2006; Amado and Monserrat, 2010; Lushchak, 2011); however, these endpoints have not been investigated following P. parvum exposure. Further, an understanding of sublethal toxicity associated with P. parvum HABs are particularly unknown across nutrient and salinity gradients known to alter HAB formation and acute toxicity.

In the present study, we examined influences of salinity and nutrients on P. parvum growth and associated acute and sublethal toxicity using two common model fish species, fathead minnow (Pimephales promelas) and zebrafish (Danio rerio). In addition to quantifying acute mortality, antioxidant gene expression, common biochemical OS biomarkers and fish photomotor behaviors were studied to determine if P. parvum elicits sublethal toxicity in fish and whether such responses varied with experimental conditions for P. parvum growth.

2. Materials and Methods

2.1. Laboratory Cultures

A Texas strain of P. parvum obtained from the University of Texas at Austin Culture Collection of Algae (UTEX strain LB 2797, Austin, TX, USA) is routinely cultured in our laboratory. A stock culture (used to inoculate experimental units) was grown at a salinity of 4 in f/2 medium at 20 °C on a 12:12 light:dark cycle (Guillard, 1975). Because this salinity condition is lower than UTEX, we cultured this strain for three months at a salinity of 4, and monitored growth and toxicity to fish prior to study initiation with medium renewals every two weeks. The stock culture was gradually acclimated, over a period of ten days (decreasing 0.5 °C per day), to 15 °C and evaluated for acute mortality to fish to confirm that the stock was acutely toxic to fish, after acclimation was complete, prior to initiating this study. A super stock of artificial seawater (ASW) was prepared by dissolving Instant Ocean salt (Spectrum Brands, Blacksburg, VA, USA) in Nanopure water (18.2 megohm ionic purity; Barnstead, ThermoFisher, Wilmington, DE, USA; Brooks et al., 2010). Media for salinity treatment levels were prepared by diluting the ASW super stock using Nanopure water to salinities of 2.4 and 5, which were chosen based on ranges commonly observed in Texas inland waters affected by P. parvum HAB events (Roelke et al. 2011, Patino et al. 2014; VanLandeghem et al. 2014a). Nutrient treatment levels were prepared by enriching water with either f/2 (882 µM NaNO₃, 36.2 µM NaH₂PO₄ H₂O) or f/8 (one fourth the concentration of f/2; 220.5 µM NaNO₃, 9.125 µM NaH₂PO₄ H₂O) nutrients, with f/2 trace metals and vitamins for both nutrient conditions (Guillard, 1975). While these nutrient concentrations are higher than reported ranges in Texas inland systems, we chose standard culture media to aid compare with previous laboratory studies (Brooks et al. 2010; Valenti et al., 2010; Israel et al., 2014; Patino et al. 2014; VanLandeghem et al. 2014b). Two replicates of each experimental unit (20 L glass carboys) were included in a replicated 2 × 2 experimental design for 8 total experimental units. ASW and all media stocks were autoclaved before use and acclimated to 15 °C. Stock cultures were introduced to each experimental unit at an initial concentration of 100 cells/mL. Experimental units were maintained in climate-controlled conditions on back-up power at 15 °C, on a 12:12 light:dark cycle at 4,000 Lux; these carboys were swirled and rotated daily. Every second day, chlorophyll a fluorescence was determined using a fluorometer (Turner Designs, San Jose, CA) to identify growth status of each culture. These subsamples collected for chlorophyll a fluorescence were preserved with 200 μL of 25% aqueous glutaraldehyde for subsequent cell counts. Cell counts were determined using a haemocytometer following previously published methods (Southard, 2005). At study initiation and stationary growth, dissolved nitrogen (nitrate/nitrite) and phosphorus (phosphate) concentrations were analyzed for each treatment level (salinity and nutrient conditions) according to standard methods using a flow-injection auto-analyzer (Lachat QuikChem 8500 and Series 520 XYZ Autosampler; APHA) (Clesceri et al. 1998). During these time points, 40 mL of each culture were sampled, filtered through a 0.45 µm filter (GN Metricel® Membrane Disc Filters, Pall Laboratory, Port Washington, NY, USA), and frozen at −20 °C in polyethylene conical tubes (ThermoFisher, Wilmington, DE, USA) following standard methods. All samples were analyzed at the same time approximately one week after study conclusion.

2.2. Aquatic Cultures

Fathead minnow (P. promelas) and zebrafish (D. rerio) used in toxicity studies were cultured at Baylor University. Fathead minnow cultures were maintained using a dechlorinated flow through system. Cultures were maintained at 25±1 °C under a 16:8 light:dark cycle. Fish were fed brine shrimp (Artemia sp. nauplii; Pentair AES, Apopka, FL, USA) twice daily. Embryos were collected from sexually mature adults aged to at least 120 d before breeding (1:4–5 male to female ratio). Larvae within 24 hours post hatch (hph) were used for acute toxicity studies and 48 hph for sublethal studies. For sublethal studies, additional <24 hph fish were acquired from Environmental Consulting and Testing (Superior, WI, USA) and incubated at 25±1 °C under a 16:8 light:dark cycle for 24 h prior to study initiation. Tropical 5D wild type zebrafish were cultured using a z-mod recirculating system (Marine Biotech Systems, Beverly, MA, USA). Fish were maintained at a density of < 4 fish per liter in 260 mg/L instant ocean with a pH of 7.0, temperature of 27±1 °C under a 16:8 light:dark cycle. Fish were fed Artemia sp. nauplii (Pentair AES, Apopka, FL, USA) with flake food (Pentair AES, Apopka, FL, USA) twice daily. Embryos used for this experiment were collected from sexually mature adults and were used for toxicity experiments at 48 hours post fertilization (hpf). All experimental procedures and fish culture protocols followed Institutional Animal Care and Use Committee protocols approved at Baylor University.

2.3. Acute Bioassays

Acute mortality bioassays using fathead minnow were employed to determine acute toxicity of cultures throughout growth stages, similar to previous published methods by our laboratory (Brooks et al., 2010; Valenti et al., 2010). Bioassays were initiated and conducted weekly when cell densities reached 10,000 cell/mL, the minimum cell density that is considered to be associated with HAB events in the field (Roelke et al., 2007). Acute mortality of fathead minnows was determined following US Environmental Protection Agency (US EPA) Toxicity Identification Evaluation (TIE) methods (US EPA 1991). Serial dilutions following a 0.5 dilution scheme were prepared for a total of six dilutions (e.g. 100, 50, 25, 12.5, 6.25, 3.125%). Media matched to experimental treatments (2.4 and 5 salinity, f/2 and f/8 nutrients) were used for dilutions and controls. Five < 24 hph fathead minnow larvae were loaded in 80 mL of each culture dilution in duplicate. pH of each solution was titrated to 8.5 using 1.0 N hydrochloric acid or 1.0 N sodium hydroxide, because increasing pH increases acute toxicity (Valenti et al., 2010). Toxicity bioassays were performed in temperature-controlled incubators at 25 °C in the dark to prevent photodegradation (James et al., 2011). Acute mortality was assessed at 48 h to estimate organism lethality (LC₅₀), which was then normalized to P. parvum cell density. Identical acute mortality studies were performed during P. parvum culture stationary growth phase with fathead minnow and zebrafish larvae (48 hph fathead minnow, 48 hpf zebrafish) to determine LC₅₀ values and subsequent sublethal study cell densities. Zebrafish bioassays were conducted in a temperature-controlled chamber set to 27 °C in the dark generally following previously reported methods (Kristofco et al., 2016; Corrales et al., 2017; Steele et al. 2018a, b).

2.4. Sublethal Bioassays

The experimental design for sublethal toxicity studies followed previously published methods by our laboratory (Corrales et al., 2017; Steele et al., 2018a, b; Steele et al., 2020). Sublethal studies with fathead minnow and zebrafish were performed according to standardized toxicity methods from the US EPA Whole Effluent Toxicity (WET) methods (EPA 2002) and Organization for Economic Cooperation and Development (OECD) Fish Embryo Toxicity Test (FET), respectively (OECD no. 236). Three dilution levels (Table 1) per experimental replicate carboy were selected below the 48 h LC₅₀ value and cell densities were targeted at which responses were observed in preliminary studies (unpublished data). This approach was taken to aid in comparisons between treatment and dilution levels. Here again, media (2.4 or 5 salinity, f/2 or f/8) was employed for dilutions and controls. Fish were acclimated to experimental media for approximately 2 hours in a 50:50 ratio of culture water: P. parvum media or reconstituted hard water (RHW) served as a negative control. Prior to fish introduction, pH of each solution was titrated to 8.5 using 1.0 N hydrochloric acid or 1.0 N sodium hydroxide to normalize pH dependent toxicity between experimental units (Valenti et al., 2010).

Table 1.

Prymnesium parvum growth rates (d⁻¹) grown at a salinity of 2.4 and 5 at 15 °C with a 12:12 light:dark cycle, under nutrient sufficient (f/2) and deficient (f/8) conditions in duplicate (represented as rep 1 and 2). Growth rates were determined when stationary growth phase was reached at study day 48 for 2.4 salinity cultures and study day 54 for 5 salinity cultures.

2.4 f/2 medium (rep 1)	2.4 f/2 medium (rep 2)	2.4 f/8 medium (rep 1)	2.4 f/8 medium (rep 2)	5 f/2 medium (rep 1)	5 f/2 medium (rep 2)	5 f/8 medium (rep 1)	5 f/8 medium (rep 2)
0.152	0.159	0.146	0.146	0.161	0.161	0.152	0.153

Sublethal bioassays were conducted in temperature-controlled chambers at 25 or 27 °C for fathead minnow and zebrafish, respectively, and in the dark, again to prevent toxin photodegradation (James et al., 2011). Briefly, 48 hph fathead minnow larvae were placed in 200 mL of experimental solutions in 500 mL glass beakers, which served as experimental units for toxicity bioassays. Sublethal assays were performed for each experimental carboy replicate (8 total), and each assay dilution level consisted of eight experimental replicates, each with ten fathead minnows. Zebrafish at 48 hpf were exposed to 30 mL of each experimental dilution in experimental units (100 mL glass beakers). Similar to the fathead minnow sublethal assays, we performed a zebrafish assay for each experimental carboy, for which each dilution consisted of twelve experimental units, each with fifteen zebrafish. These volumes were chosen to ensure that the loading density did not exceed acceptable levels for standardized guidelines. Here again, this general experimental design followed our previously reported work (Corrales et al., 2017). Fish mortality and abnormalities were observed and recorded at 24 and 48 h.

At 48 h, fish larvae were collected, frozen at −80 °C, and saved for antioxidant related gene expression and biochemical OS biomarker determination (Corrales et al. 2017; Steele et al., 2020), and additional fathead minnow and zebrafish larvae photomotor behaviors were observed (Steele et al., 2018a, b; Steele et al., 2020). Five and ten fathead minnow larvae were pooled per experimental unit for real time reverse transcription polymerase chain reaction (qPCR) gene expression and OS biochemical determination, respectively. A total of three experimental units were collected for both molecular and biochemical endpoints. Four fathead minnow were analyzed individually for behavior from each experimental unit. A total of three experimental units were randomly selected for behavioral analysis. In the case of zebrafish (due to smaller size of the fish), fifteen and ten zebrafish larvae were pooled per beaker for OS biochemical and qPCR gene expression, respectively. Similar to fathead minnow, a total of three experimental units were collected per endpoint. If no mortalities occurred, remaining tissue were collected for additional DNA and RNA extractions. Six zebrafish were analyzed individually for behavior from one experimental unit. A total of four experimental units were randomly selected for behavioral analysis. Each collection of organisms per experimental unit represented a replicate. Therefore, a total number of three replicates were statistically analyzed for all endpoints for each dilution of samples from each experimental carboy. Zebrafish behavior consisted of four replicates that were statistically analyzed following previously published methods from our laboratory (Steele et al., 2018a, b; Steele et al., 2020).

2.5. Antioxidant Gene Expression

Changes in mRNA abundance were measured for glutamate cysteine ligase catalytic subunit (gclc), glutathione-S-transferase (gst), nuclear factor erythroid-2 like 2 (nrf2), and superoxide dismutase (sod) in order to determine changes in genomic activity associated with OS. Specific isoforms of gst, nrf2 and sod genes measured in zebrafish were gstp1, nrf 2a, and sod1, though due to the poor annotation of this gene family in fathead minnows we were not able to identify the specific isoforms in this species. Gene expression was determined following previously described methods (Corrales et al., 2017). RNA was extracted from whole larval fish using RNAzol (Molecular Research Center, Cincinnati, OH, USA) and cleaned and purified using RNeasy Mini kit (Qiagen, Valencia, CA, USA). Total RNA was measured using a Nanodrop One^C (Thermo Scientific, Wilmington, DE, USA), and 500 ng of total RNA was reversed transcribed to cDNA using TaqMan Reverse Transcription reagents (Applied Biosystems by Life Technologies, Carlsbad, CA, USA) to yield 25 ng/µL reaction.

Relative abundance of target genes was determined by real time reverse transcription polymerase chain reaction (qPCR). This reaction consisted of 1 µL cDNA, 300 nM of each forward and reverse primer, and 1X Power SYBR Green PCR Master Mix. Gene amplification reaction conditions were 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 s, and 60 °C for 1 min using a StepOnePlus Real-Time PCR System (Applied Biosystems by Life Technologies, Carlsbad, CA, USA). Reaction of each sample was performed with two technical replicates per biological replicate (triplicate). Prior to performing assays, amplification efficiencies of all primer pairs were determined at ≥90%. Beta-actin (actb), glyceraldehyde-3-phosphate dehydrogenase (gapdh), and hypoxanthine phosphoribosyltransferase 1 (hprt1) were selected as reference genes. The geometric mean of actb1 and gapdh for zebrafish and that of gapdh and hprt1 for fathead minnow were used as controls to normalize the starting quantity of mRNA in target genes.

2.6. Biochemical Oxidative Stress Biomarkers

Determination of total glutathione concentration, lipid peroxidation, and oxidative DNA damage also followed previously described methods (Corrales et al., 2017). Briefly, total glutathione (GSH) concentration was determined using a commercially available kit (Cayman Chemical Company, Ann Arbor, MI, USA). Prior to conducting the assay, samples were deproteinated with 1.25 M metaphosphoric acid and 0.2 M triethanolamine. DTNB (5,5,-dithio-bis-2-nitrobenzoic acid) was added to deproteinated tissue supernatant initiating a reaction between the GSH present in tissue samples and DTNB yielding TNB (5-thio-2-nitrobenzoic acid). The rate of TNB production is directly proportional to the GSH concentration due to the recycling of GSH by glutathione reductase present. Total glutathione concentrations were normalized to sample protein content. Protein content was determined following the Bradford protein assay by which a Bio-Rad protein dye was reacted with tissue supernatant (Sigma-Aldrich, St. Louis, MO, USA Cat. No. A7906 and 5000006).

Lipid peroxidation was determined by the concentration of malondialdehyde (MDA) present in fish tissue samples. MDA is a reactive carbonyl compound that is a natural product of lipid peroxidation. MDA concentration was quantified using a commercially available Thiobarbituric Acid Reactive Substances assay kit (TBARS) (Cayman Chemical Company). Thiobarbituric acid (TBA) was added to each tissue sample, producing a MDA-TBA adduct which was fluorometrically detected. Elevated MDA concentrations are proportional to the MDA-TBA adducts formed. MDA concentration was also normalized to sample protein content. Protein content was determined following the Pierce BSA assay by which a working dye reagent was reacted with tissue supernatant (Thermo Scientific Wilmington, DE, USA, Cat No. 23225).

Oxidative DNA damage was determined by presence of the oxidatively damaged guanine species, 8-hydroxy-2’-deoxyguanosine (8-OH-dG), measured using a commercially available enzyme immunoassay (EIA) (Cayman Chemical Company). Prior to conducting EIA, DNA was extracted using DNAzol (Molecular Research Center, Cincinnati, OH, USA) following the manufacturer’s instructions. Extracted DNA samples were cleaned and purified using Zymo Genomic DNA Clean and Concentrator (Zymo Research, Irvine, CA, USA) prior to DNA quantification. DNA concentrations were quantified using a Nanodrop One^C (Thermo Scientific), and 5 μg DNA per sample was prepared for the EIA by diluting DNA with Cayman Ultrapure water to yield a 50 μg/mL sample. The amount of 8-OH-dG present in the sample competed with an added 8-OH-dG-acetylcholinesterase conjugate for binding to an oxidative damage monoclonal antibody. This antibody was bound to seeded goat polyclonal antimouse IgG cells. After an 18 h incubation, each plate was washed five times and Ellman’s reagent was added to develop the plate. The intensity of the signal is inversely proportional to the amount of free 8-OH-dG or oxidatively damaged DNA.

2.7. Behavioral Analyses

Behavioral responses of fathead minnow and zebrafish were observed after 48 h sublethal exposures to P. parvum following previously described methods (Kristofco et al., 2016; Steele et al., 2018a, b; Steele et al., 2020). All well plates were preloaded and maintained in exposure conditions until analysis. Malformed fish were excluded and each dilution level was represented on a plate. Fish were acclimated to the well plate for at least 30 minutes before behavioral platform loading. To minimize time of day behavioral effects, plates were analyzed from approximately 9:00 am to 2:00 pm for fathead minnow and 2:00–7:00 pm for zebrafish with each plate analyzed immediately after the conclusion of previous plate (Kristofco et al., 2016).

Larval swimming patterns were observed and recorded using automated tracking software (ViewPoint, Lyon, France) and associated platform (Zebrabox, ViewPoint, Lyon France). This system was set in tracking mode and behavioral recordings took place over 50 minutes with a ten minute dark acclimation period followed by two altering ten minute light/dark cycles. Observations were recorded for total distance swam and total number of movements. Additionally, distance swam, number of movements, and duration of movements were recorded for activity across three different speed thresholds. These speeds are categorized as bursting (>20 mm/s), cruising (5–20 mm/s), and freezing (<5 mm/s) to characterize stimulatory and refractory behaviors. To measure larvae swimming responses to a sudden change in photoperiod condition, photomotor response (PMR) was calculated following methods previously used (van Woudenberg et al., 2013) with minor modifications (Steele et al., 2018a, b; Steele et al., 2020). PMR for each photoperiod transition (two light and two dark responses) was calculated as the change in mean distance traveled (mm) between the last minute of an initial photoperiod and the first minute of the following period.

2.8. Statistical Analyses

P. parvum specific growth rates were calculated using the equation:

$$\text{r}=\left(\ln {\text{N}}_{\text{d}}-\ln {\text{N}}_0\right)/\text{t}$$

where r is the growth rate (d⁻¹), N_d is the number of organisms at the beginning of the steady growth state, N₀ is the number of organisms at study initiation, and t is the time (days) to reach steady state growth. Steady state growth was determined as the time at which the maximum P. parvum density was reached and followed by a general decline.

The lethal concentration causing 50% mortality (LC₅₀ values) from acute studies with each of the cultures were calculated using the Toxicity Relationship Analysis Program version 1.30a (EPA). Sigma Plot 13.0 (Systat Software Inc., San Jose, CA, USA) software was used for statistical analyses of P. parvum growth and antioxidant gene expression, biochemical markers, and fish behavior. Whether significant mortalities (α=0.05) from control at 48 h occurred was tested for all treatment dilutions selected for the sublethal bioassays using a Fisher Exact test (US EPA 2002). All sublethal responses were not measured for treatment dilutions with significant mortalities compared to the media control. Prior to analysis antioxidant gene expression data were normalized to the geometric mean of reference/endogenous genes and then normalized to media controls to determine linearized 2−ΔCt values, prior to normality and equivalence of variance analyses. Biochemical and behavioral data were normalized to cell density (cells/mL) and then assessed for normality and equivalence of variance.

Significant differences (α=0.05) in mRNA fold changes, biochemical OS endpoints, and behavior movement patterns were identified among dilution levels from each experimental replicate carboy using a One-Way ANOVA if normality and variance assumptions were met. Dunnett’s post hoc test was performed to identify dilution level differences from media controls and used to derive No Observable Effect Concentration (NOEC) and Lowest Observable Effect Concentration (LOEC). For data not meeting ANOVA assumptions, an ANOVA on ranks was performed. NOEC values were log transformed prior to statistical analysis to meet normality assumptions. NOEC values less than the lowest dilution level selected were excluded in statistical analysis. Significant differences (α=0.05) of maximal cell densities and growth rates were identified for main and interacting treatment factors (salinity, nutrients, salinity x nutrients) for NOEC values determined for each experimental carboy replicate using General Linear Models (GLM) with SPSS software (IBM Corp., Armonk, NY, USA). Main and interacting treatment factors (salinity, nutrients, salinity x nutrients) effects were also determined with GLMs for NOEC values for OS related endpoints and behavior.

3. Results

3.1. P. parvum Growth

In the present study, specific growth rate of experimental units ranged from 0.146 – 0.161 d⁻¹ (Table 1). Cultures grown under the same treatment conditions, except low salinity high nutrient, were not significantly different (p > 0.05) but growth was influenced by salinity and nutrient factors (p<0.05) (Table S1). Specifically, nutrient limitation resulted in lower growth rates and lower maximal cell densities at study conclusion, compared to nutrient sufficient conditions, regardless of salinity (Figure 1). The low salinity (2.4), nutrient sufficient cultures were comparable to those grown under higher (5) salinity with deficient nutrients. Maximal cell densities also varied with salinity and nutrient condition and were significantly (p < 0.05) influenced by main and interacting growth conditions (Table S1). Higher salinity and nutrient sufficient conditions resulted in the highest cell densities observed, followed by higher salinity with deficient nutrients, low salinity with sufficient nutrients, and low salinity with nutrient deficiency. Exponential growth began on day 26 of the study for all experimental culture conditions. Stationary growth was reached at day 48 for 2.4 salinity cultures and day 54 for 5 salinity cultures.

Figure 1:

Growth of Prymnesium parvum represented as mean cell density (cells/mL, n=5) · S.D. per study day. Cultures were grown in duplicate at A) 2.4 and B) 5 salinity, 15 C with a 12:12 light:dark cycle, under nutrient sufficient (f/2) and deficient (f/8) conditions. Cell densities were determined using a haemocytometer. Acute toxicity tests were initiated every 7 days after exponential growth phase began. Larval fathead minnow (Pimephales promelas) LC50 values (P. parvum cell density) are represented by bars ±95% C.I. Nontoxic cultures represented by #. Dotted line represents a bloom threshold (10,000 cells/mL).

3.2. Acute Bioassays and Sublethal Survival

Larval fathead minnow LC₅₀ values indicated that all cultures were highly toxic throughout the duration of P. parvum growth (Figure 1). Fathead minnow and zebrafish treatment control survival was not affected by either salinity and nutrient conditions (survival >90%). Acute mortality was observed at the start of exponential growth and continued to stationary growth phase (Table S2). Acute mortality of fathead minnow decreased with study day (Figure 1 and Table S2) for all cultures, with the highest mortality occurring during early exponential growth phase. Acute toxicity to fish was highest under nutrient deficient conditions for both salinities. However, acute mortality was greatest under higher salinity and deficient nutrient conditions (Figure 1). Zebrafish exhibited biphasic toxicity after exposure to the high salinity cultures, with greatest mortality between 37,200–292,000 cells/mL (Table S3). Similarly, survival was greatest under nutrient sufficient, compared to nutrient deficient, conditions during the sublethal bioassays for both species (Figure S1). Significant (p < 0.05) differences in mortality were observed for the highest dilution levels selected for experimental carboy replicates that were acutely toxic prior to sublethal exposures; these dilution levels were omitted in sublethal response evaluations (Figure S1).

3.3. Antioxidant Gene Expression

Relative gene expression (mRNA fold change) was significantly (p < 0.05) induced, unless otherwise stated, in both species, but varied among P. parvum culture conditions. Fathead minnow gclc and nrf2 induction (p<0.05) occurred after exposure to the higher salinity treatment level (Figures 2 & S2). A low salinity x deficient nutrient culture also significantly (p<0.05) induced nrf2 expression (Figure 2.6). Fathead nrf2 NOEC values were observed at the lowest cell densities and comparable among cultures that consistently elicited acute mortality prior to sublethal study (Tables S2 & S5). For example, 5 salinity f/2 rep 1 and both f/8 cultures induced nrf2 expression in fathead minnow larvae at similar cell densities and were toxic during exponential and stationary growth of P. parvum. Higher salinity and nutrient sufficient conditions induced gclc expression in zebrafish, which was significantly (p<0.05) influenced by salinity (Figures S2, Table S6). Significant (p<0.05) inductions of nrf2a were also observed after exposure to this same experimental treatment condition (Figure 2). Comparable to GSH and lipid peroxidation, nrf2 expression in fathead minnow was significantly (p<0.05) influenced by nutrient condition and interactive effects (Table S6), whereas nrf2a expression was only influenced by nutrients (Table S6). Though not significant (p > 0.05), large inductions of gclc and gst were observed for fathead minnow when exposed to nutrient deficient cultures, while zebrafish conversely elicited depletions (Figures S2–3). Similar to fathead minnows, nrf2a expression was induced after exposure to the same low salinity x deficient nutrient culture (Figure 2). Higher salinity with deficient nutrients resulted in a significant (p<0.05) depletion of gstp1 in zebrafish (Figure S3). No significant differences were observed for sod/sod1 expression for either species; however, significant influences of salinity and nutrients were observed for zebrafish NOEC values (Figure S4, Table S6).

Figure 2:

Mean · S.E relative nrf2 gene expression (mRNA fold induction change) determined by qPCR in larval fathead minnow (Pimephales promelas; A, C) and zebrafish (Danio rerio; B,D) after 48 h exposure to Prymnesium parvum cultures grown under 2.4 (A-B) or 5 (C-D) salinity and nutrient sufficient (high, f/2) or deficient (low, f/8) conditions. Data were first normalized to reference genes and then compared to control. Statistical significance was determined using a one-way ANOVA followed Dunnett’s post hoc test (*: p ≤ 0.05, N=3). Media control (0 cells/mL) contained no P. parvum cells.

3.4. Biochemical Oxidative Stress Endpoints

Fathead minnow and zebrafish exhibited differential biochemical OS responses across nutrient and salinity treatments. Statistically significant (p < 0.05) depletion of total glutathione and increases in lipid peroxidation were observed in fathead minnow exposed to low salinity and nutrient deficient conditions with similar trends after exposure to remaining culture conditions (Figures S5–6). Both of these responses were significantly (p < 0.05) influenced by nutrients and an interaction between salinity and nutrients (Table S6). NOECs were comparable among cultures that elicited acute fathead minnow mortality with the most significant responses observed after exposure to low salinity and low nutrient conditions (Table 2). Increases in oxidative DNA damage were observed after exposure to higher salinity cultures; however, these responses were not significant (Figure S7). By contrast, zebrafish elicited a decrease in lipid peroxidation after exposure to higher salinity and nutrient sufficient conditions (Figure S6). Higher salinity and nutrient deficiency significantly decreased oxidative DNA damage in zebrafish (Figure S7). Though not significant (p > 0.05), higher salinity with nutrient deficient treatment increased total glutathione concentration in zebrafish (Figure S5). GSH, lipid peroxidation and DNA damage in zebrafish were significantly (p < 0.05) influenced by salinity and nutrient conditions (Table S6). No significant glutathione depletion was observed as indicated by NOEC values; therefore, this significant influence could be controlled by P. parvum cell density (Table 2). Lipid peroxidation and DNA damage were observed at the lowest dilution; thus, a NOEC value could not be determined (Table 2).

Table 2.

No Observed Effect Concentration (NOEC) and Lowest Observed Effect Concentration (LOEC) (Prymnesium parvum cells/mL) values to elicit total glutathione (GSH), lipid peroxidation (LPO) and DNA damage responses in larval fathead minnow and zebrafish following a 48 h exposure to P. parvum grown under varying salinity (2.4 or 5) and nutrient conditions (nutrient sufficient; f/2 or deficient; f/8) in duplicate (represented as rep 1 and 2).

	Fathead minnow						Zebrafish
	GSH		LPO		DNA damage		GSH		LPO		DNA damage
Culture condition:	NOEC	LOEC	NOEC	LOEC	NOEC	LOEC	NOEC	LOEC	NOEC	LOEC	NOEC	LOEC
2.4, f/2 medium (rep 1)	38,500	>38,500	38,500	>38,500	38,500	>38,500	38,500	>38,500	38,500	>38,500	38,500	>38,500
2.4, f/2 medium (rep 2)	21,200	>21,200	21,200	>21,200	21,200	>21,200	53,000	>53,000	53,000	>53,000	53,000	>53,000
2.4, f/8 medium (rep 1)	1,240	12,400 (↑)	1,240	12,400 (↑)	12,400	>12,400	31,000	>31,000	31,000	>31,000	31,000	>31,000
2.4, f/8 medium (rep 2)	1,080	>1,080	1,080	>1,080	1,080	>1,080	27,000	>27,000	27,000	>27,000	27,000	>27,000
5, f/2 medium (rep 1)	11,600	>11,600	11,600	>11,600	11,600	>11,600	23,200	>23,200	23,200	>23,200	23,200	>23,200
5, f/2 medium (rep 2)	4,670	>4,670	4,670	>4,670	4,670	>4,670	23,700	>23,700	<11,700	11,700 (↓)	23,400	>23,400
5, f/8 medium (rep 1)	7,480	>7,480	7,480	>7,480	7,480	>7,480	15,000	>15,000	15,000	>15,000	<7,480	7,480 (↓)
5, f/8 medium (rep 2)	2,980	>2,806	2,806	>2,806	2,806	>2,806	14,900	>14,900	14,900	>14,900	14,900	>14,900

N=3 (p ≤ 0.05, induction or depletion represented by ↑ ↓).

3.5. Fish Behavioral Effects

Fish photomotor behavioral responses to P. parvum varied among species and experimental culture conditions. Fathead minnow activity was greatest in light conditions, indicated by the positive dark to light PMRs and reduced light to dark PMRs (Figures 3, S8–10). Statistically significant (p < 0.05) decreases in swimming activity occurred in the light and dark conditions for low salinity with sufficient nutrients and high salinity with deficient nutrients, with responses observed at lower cell densities when grown under higher salinity conditions (Figures 3, S8). Though decreased activity was significantly (p < 0.05) affected across freezing and cruising speed thresholds, stimulatory response trends were interestingly observed for bursting speed thresholds (Figure 3, S8–10). An increase in activity was observed for fathead minnow exposed to higher salinity conditions with sufficient nutrients (Figure S10). Stimulatory responses across bursting thresholds were observed for fathead minnow exposed to each culture condition, however this trend was not observed in all replicates. Salinity and nutrient conditions, and their interaction, did not significantly (p > 0.05) influence PMR responses (Tables S7–8).

Figure 3:

Fathead minnow (Pimephales promelas) photomotor swimming responses (A, C) following a sudden change in light condition (Light to Dark and Dark to Light) and swimming activity (B,D) during light and dark conditions after 48 h exposure to Prymnesium parvum grown in duplicate (rep 1 A,B; rep 2 C,D) at 5 salinity under nutrient deficient (f/8) conditions. Swimming data depicts distance swam, total number of movements (counts), and duration of each movement across 3 speed levels: freezing (<5 mm/sec), cruising (5–20 mm/sec), and bursting (>20 mm/sec). Data were normalized to control (red line), representing 0 on the axis. * and arrows = p ≤ 0.05 (N=3) increased (↑) or decreased (↓) behavioral response.

Zebrafish activity was greatest in dark conditions indicated by the negative dark to light and positive light to dark PMRs (Figures 4, S11–13). A statistically significant (p < 0.05) increase in PMR was observed under higher salinity and nutrient limited conditions (Figures 4). Low salinity with sufficient nutrients elicited significant (p < 0.05) stimulatory responses in zebrafish swimming behavior in the light (Figure S11). Similar to the fathead minnow, stimulatory trends were observed for bursting speed threshold with heightened responses in light conditions, although some decreases in activity were observed (Figures 4, S11–13). Contrary to fathead minnow, salinity and nutrients did influence swimming behavior and PMR zebrafish responses (Tables S7–8). Though there were significant (p < 0.05) influences of both these experimental factors, it should be noted that these NOEC values were the highest dilution level exposed to zebrafish, except for light to dark PMRs.

Figure 4:

Zebrafish (Danio rerio) photomotor swimming responses (A, C) following a sudden change in light condition (Light to Dark and Dark to Light) and swimming activity (B,D) during light and dark conditions after 48 h exposure to Prymnesium parvum grown in duplicate (rep 1 A,B; rep 2 C,D) at 5 salinity under nutrient sufficient (f/2) conditions. Swimming data depicts distance swam, total number of movements (counts), and duration of each movement across 3 speed levels: freezing (<5 mm/sec), cruising (5–20 mm/sec), and bursting (>20 mm/sec). Data were normalized to control (red line), representing 0 on the axis. * and arrows = p ≤ 0.05 (N=4) increased (↑) or decreased (↓) behavioral response.

4. Discussion

P. parvum HABs are characterized by appreciable fish mortality events that apparently target gill breathing organisms. These HAB events are expected to continue to increase due to climate change, watershed modifications and urbanization. National and international scale monitoring networks and databases are relatively nonexistent for P. parvum, signaling concern due to the projected increase and expansion of these HABs. Herein, understanding aquatic impacts of P. parvum blooms can assist with prioritization of susceptible systems. In the present study, a novel approach was taken to understand sublethal P. parvum toxicity by examining multiple molecular, biochemical and behavioral responses. The potential influence of nutrients and salinity were investigated to understand these effects at conditions representative of Texas inland HABs. Thus, we focused on the Texas P. parvum strain (LB 2797) as an initial exploration of potential sublethal perturbations elicited to fish. To our knowledge, this is the first study to report induced OS biomarker and behavioral alterations in fish following exposure to sublethal cell densities of P. parvum. We found that these responses were heightened under suboptimal growth conditions, which is consistent with previously reported acute mortality (Baker et al., 2007, 2009; Valenti et al., 2010). Similar sublethal approaches should be applied to other P. parvum strains in the future given the current results and differential acute mortality observed previously among strains (Blossom et al. 2014b).

P. parvum is a relatively slow growing species with optimal growth rates between 0.8–1.8 d⁻¹ (Baker et al., 2007) that vary based on strain and associated optimal growth conditions (Blossom et al. 2014b). Under stressful conditions representative of inland Texas HABs (including nutrient limitation, lower salinity, and winter temperatures), reproductive growth rates have been shown to decrease to around 0.1–0.3 d⁻¹ (Baker et al., 2009). The growth rates observed in the present study indicate that P. parvum growth was not optimal for any of the experimental conditions (Table 1) and are comparable to Texas HABs (Baker et al., 2007; Baker et al., 2009). Suboptimal growth conditions such as low salinity, temperature, and nutrients have been identified as indicators of P. parvum bloom formation and acute toxicity (Brooks et al., 2011; Roelke et al., 2016). However, previous laboratory studies have seldom compared multiple salinity and nutrient conditions concurrently (Brooks et al., 2010; Flood and Burkholder, 2018).

Salinity particularly influences P. parvum in which growth can occur above salinity of 1, and lower salinity conditions apparently stress the organism, resulting in toxins production and elevated acute toxicity (Roelke et al., 2016). Other factors such as light intensity, stemming from seasonal changes in the field and a lack of standardized laboratory culturing procedures for P. parvum, can influence P. parvum growth and contribute to differential toxicity. We maintained light conditions throughout our study to avoid light being a confounding factor for interpretation of our results. A relationship between temperature and low salinity conditions exist in which the optimal temperature for growth decreases with decreasing salinity (Baker et al., 2009). In the current study, P. parvum growth (maximal cell density and growth rate) was greatest under the higher salinity treatment level. Slight increases in salinity have been shown to increase P. parvum blooms in South Central U.S.A. (Roelke et al., 2011; Hambright et al., 2014), which are consistent with observations in the present laboratory study. In addition, we observed lower maximal cell densities and growth rates and an increase in acute fish mortality in experimental units grown under nutrient limitation, regardless of salinity. Similar observations were made for sublethal responses. In fact, nutrient limitation significantly and most frequently influenced the sublethal endpoints examined here (Tables S6–8).

Though analytical determination of prymnesins is rare, highest concentrations of prymnesins have been reported throughout late exponential and stationary growth for cultures grown under similar conditions as the present study (La Claire et al., 2015). Additionally, the type of prymnesin(s) detected has shown to vary by P. parvum strain (Binzer et al. 2019). As introduced above, no analytical standards exist for prymnesin I and II or other prymnesins. Therefore, acute bioassays were employed in the current study to provide a surrogate measure for the presence of bioavailable toxins. Low salinity and nutrient sufficient conditions became nontoxic during late stationary phase; whether prymnesins or other substances were causative toxins responsible for acute fish mortality is not known (Figure 1, Table S2). In fact, fathead minnow mortality decreased per study day with the highest toxicity per cell observed during early exponential phase for all cultures (Figure 1). Svenssen et al. (2019) reported highest concentration of prymnesins (B-type) during exponential growth phase with declines of these toxins during late exponential and stationary phase, which generally supports our mortality observations despite different culture conditions and strains utilized. Multiple substances and varieties of prymnesins have also been suggested to be the toxins responsible for fish mortalities including fatty acids, fatty acid amides, and other golden algae toxins (GATs), though there is much discrepancy among these results (Henrikson et al.,2010; Schug et al., 2010; Bertin et al., 2012a; Bertin et al., 2012b; Blossom et al, 2014a, Blossom et al, 2014b; Hems et al. 2018; Binzer et al., 2019; Svenssen et al. 2019). A plausible explanation for such differences among proposed toxins could be influenced by differential toxins production and/or the magnitude of toxins produced under different growth conditions, during different growth phases, reported toxicity responses, and/or a combination of these factors. Future investigations are required to better understand mechanisms contributing to toxins production and resulting toxicity.

Fathead minnow and zebrafish were differentially acutely sensitive to P. parvum. In the present work, fathead minnow mortality was more sensitive to P. parvum cultures than zebrafish (Table S2–3). Though the mechanisms behind these differences were not explored here or elsewhere, the different developmental stages employed could have influenced the observed mortality responses. For example, increased chemical uptake and metabolism with development have been demonstrated for multiple fish species exposed to environmental contaminants including algal toxins (Wiegand et al., 1999, 2000; Otte et al., 2010; Kristofco et al., 2018). For example, we recently (Kristofco et al., 2018) identified that older zebrafish larvae were more sensitive to an environmental contaminant than embryos and younger larvae, which is consistent with results of the present study because fathead minnows were older and more developed than zebrafish. Additionally, basal antioxidant gene expression has also been shown to change with zebrafish developmental stage with higher expression occurring immediately after hatching (Wiegand et al., 2000). In the present study, we specifically employed standardized toxicological methods for the fathead minnow and zebrafish from the US EPA and the OECD, respectively, to maximize comparability of our findings to other studies. These methods are initiated with organisms at different stages of development. Whether the differences between fish responses observed in the present study are due to toxicodynamic or toxicokinetic differences originating from either different species, developmental stages or influences of both warrants further investigation, particularly because there is limited information on zebrafish responses to P. parvum, making direct comparisons among species difficult at this time. Fathead minnow mortality was elevated by nutrient limitation at both salinity treatment levels, which is consistent with previously reported results (Graneli and Johansson, 2003; Errera et al., 2008; Baker et al., 2009; Valenti et al, 2010). When P. parvum grown at 2.4 salinity under the same nutrient limited conditions (f/8), Brooks et al. (2010) reported a mean LC₅₀ value of 21,800 cells/mL for larval fathead minnows during stationary growth phase, which is quite similar to 21,725 cells/mL (Table S2), an experimental replicate of the low salinity (2.4) and nutrient deficient treatment level during stationary growth phase in the present study. Mortality of 10–14 day old fathead minnow following exposure to P. parvum grown at 6 salinity resulted in a mean LC₅₀ value of 51,560 cells/mL during stationary growth (Remmel and Hambright, 2012). As noted by Brooks et al. (2010), development and application of standardized growth and toxicity procedures are required in order to adequately compare reported toxicity within and across studies.

Oxidative stress (OS) is a common component in any substantial stress that results in an imbalance between reactive oxygen species (ROS) and antioxidant capabilities to detoxify these molecules. Because exposure to increased ROS may fluctuate, organisms are able to respond to this stress by inducing antioxidant related enzymes. The Keap1-Nrf2 signaling pathway is important in regulating antioxidant enzymes in response to xenobiotics that is conserved in vertebrate systems (Kaspar et al., 2009; Nguyen et al., 2009; Lushchak, 2011). Nrf2 is a leucine zipper transcription factor that under normal physiological conditions is complexed with the repressor protein Keap1 (Kelch-like ECH associating protein 1) in the cytosol. Cell exposure to oxidants changes the confirmation of Keap1, leading to the release of Nrf2, which migrates to the nucleus and binds to the antioxidant response element (ARE) activating gene transcription (Lushchak, 2011). The present study evaluated the expression changes of four genes after exposure to P. parvum: glutamate cysteine ligase catalytic subunit (gclc), glutathione-s-transferase (gst), nuclear factor erythroid 2–like 2 (nrf2), and superoxide dismutase (sod). The specific gene isoforms for zebrafish were gstp1, nrf2a and sod1. These four specific gene expression changes have been reported for mammalian (mice) and fish following other algal exposures, primarily cyanobacteria (Jos et al., 2005; Wang et al., 2006; Qiu et al., 2007; Gonçalves-Soares et al., 2012; Dorantes-Aranda et al., 2015), while gclc, gst, and nrf2 remain unstudied after exposure to P. parvum or other algal toxins.

Relative gene expression was widely induced across both species (Fig. 2, S2, S3, and S4), highlighting the sensitivity of these antioxidants in response to OS elicited by P. parvum. For fathead minnow, nrf2 expression was the most sensitive antioxidant molecular response examined. This is not surprising due to the major role that the Nrf2-Keap1 pathway plays in OS defense. The second most sensitive antioxidant gene was gclc for both species. Glutamate cysteine ligase (gcl) consists of a catalytic (gclc) and a light or modifier subunit (gclm) that is involved in the first step of GSH synthesis. Gclc is the rate limiting step for GSH synthesis thus expression is upregulated when increased cellular defenses are needed but if insults persist may become dysregulated (Lu, 2013). Another GSH related enzyme, glutathione-s-transferase (gst) was induced by P. parvum in zebrafish. Gst facilitates the conjugation between GSH and a reactive molecule increasing hydrophilicity for excretion. Induction of gst expression has occurred in response to OS induced by environmental contaminants and algal toxins (Limon-Pacheco and Consebatt, 2009). Superoxide dismutase (sod) is responsible for the partitioning of superoxide radicals into hydrogen peroxide and molecular oxygen that is often induced in aquatic organisms as a result of OS (Di Giulio et al., 1989). No significant responses were observed for sod in species (Fig. S4), which was similarly observed in in vitro investigations of P. parvum exposure to gill cells (Dorantes-Aranda et al., 2015). Consistent with the higher mortality sensitivity observed in fathead minnow, significant changes in gene expression were more commonly observed for fathead minnow than zebrafish, especially under conditions at the higher salinity treatment level.

Implications of OS toxicity include damage to tissues, inflammation, carcinogenesis and neurodegenerative diseases in humans and wildlife (Scandalios, 2005; Kensler et al., 2007). Gill cytotoxicity leading to tissue damage has been associated with P. parvum exposure and currently is a major tissue targeted by prymnesins (Shilo et al. 1967). Gill mucus secretion and lesions have been observed for adult fish during an active bloom, however, we did not observe any lesions, secretion, or tissue damage on larvae exposed to any of the P. parvum cultures (Lindhom et al. 1999; Vasas et al., 2012). In addition, micropredation, as a result of the attachment of P. parvum cells to zooplankton (Daphnia) and fathead minnow gill tissues, was previously reported to correlate with mortality (Remmel and Hambright, 2012). No attachment of P. parvum cells were observed on either fish species exposed to any of the P. parvum experimental conditions; however, our assessment of larvae occurred rapidly to minimize the amount of light exposure, which can apparently lead to photolysis of causative toxins (James et al., 2011). Due to the small size of fish larvae, whole fish homogenates are commonly evaluated for biochemical and molecular OS biomarkers during aquatic toxicology studies. Specific tissue-level responses in combination with tissue damage (e.g., histopathology) should be performed to further increase an understanding of P. parvum sublethal toxicity.

The main targets of ROS include proteins, lipids, and nucleic acids therefore the alterations of glutathione concentrations, lipid peroxidation, and DNA damage have been well studied biomarkers of OS (Di Giulio et al., 1989; Valavanidis et al., 2006; Limon-Pacheco and Consebatt, 2009; Lushchak, 2011). In the present study, all fish OS biochemical markers responded similarly, with neither more sensitive than the others. GSH is a tripeptide that serves as a cosubstrate during phase II metabolism for xenobiotic detoxification and is an essential electron donor for the reduction of hydroperoxides, by serving as an electron donor to glutathione peroxidases. GSH is located largely in the cytosol primarily in its reduced form and is oxidized into glutathione disulfide (GSSG) upon contact with electrophilic compounds. An increase in GSSG leads to a depletion of cellular GSH, suggesting oxidative stress and other pathological conditions (Wu et al., 2004). Therefore, the GSH depletion observed for larval zebrafish and fathead minnow is indicative of oxidative stress and suggests that GSH is actively involved in detoxification or decreased synthesis in response to exposure to P. parvum toxins.

Lipid peroxidation often leads to impairment of cell membranes. MDA has been a commonly studied end product of lipid peroxidation that can react with biomolecules forming adducts. These adducts can then undergo secondary reactions with DNA and proteins, altering properties of biomolecules and accumulate during aging and chronic diseases (Ayala et al., 2014). Consequently, an increase in lipid peroxidation is suggestive of OS as observed in the present study by the increase in MDA concentration in fathead minnow. Interestingly, we observed a decrease of lipid peroxidation in zebrafish, indicating an antioxidant response rather than an OS elicited effect. A possible explanation for this decline is metabolism of MDA and detoxification as supported by an induction of nrf2a observed after exposure to the same culture (Ayala et al., 2014). Lipid peroxidation was induced within 12 h of exposure to microcystin-LR but decreased to control levels after 24 h in zebrafish brain (Zhang et al., 2013). This recovery, although incomplete due to induction of antioxidant related genes at 24 h, suggests that these responses occur rapidly. A similar increase and decrease in lipid peroxidation corresponding with an induction of an antioxidant enzyme was observed in mice (Gehringer et al., 2004).

8-hydroxy-2’-deoxyguanosine (8-OH-dG) is the most common oxidative DNA damage product that has been studied as an indicator of oxidative stress and carcinogenesis. 8-OH-dG was analyzed in conjunction with other biological markers to confirm that an increase in 8-OH-dG is accurately representative of oxidative stress (Kasai, 1997). In the present study, a significant decrease in DNA damage was observed in zebrafish, but not fathead minnow. Similar to the lipid peroxidation results, a decrease in DNA damage was identified in the higher salinity treatment level, which suggests an antioxidant response occurred after exposure to P. parvum. Accumulation of oxidized purines were observed in human hepatoma HepG2 cells after exposure to microcystin-LR (Zegura et al., 2004). Other studies have also shown the role of pretreatment protection (including GSH) against microcystin-LR DNA damage; however, DNA damage still occurred in pretreated individuals but to a lesser extent (Lakshmana Rao and Bhattacharya, 1996). In the present study, experimental treatments that exhibited highest DNA damage did not have significantly different GSH content compared to controls. However, the cultures with the highest GSH content, exhibited the lowest DNA oxidative damage when comparing across culture conditions, suggesting similar GSH protection. Here again, whether prymnesins or different secondary metabolites elicited such diverse OS biomarker responses is not known.

Novel analyses were also performed to explore whether P. parvum may alter photolocomotor behaviors of fish. Behavior represents an organism’s adaptable response to internal (physiological) and external (social) factors and is essential for survival (Gerhardt, 2007, Amiard-Triquet, C., 2009). Therefore, alterations to behavior after exposure to algal toxins may result in adverse outcomes at the individual and population levels of biological organization. Observation of adult fish behavioral changes after exposure to P. parvum has been documented in the field, in which fish were reportedly lethargic, swam with coordination problems, slow movement, and remained at the surface of the water gasping for air (Vasas et al., 2012). Such observations are important because behavioral responses to other environmental contaminants, including algal toxins, have been observed at levels orders of magnitude below which elicit effects on acute mortality (Lefebvre et al., 2004; Valenti et al., 2012; Zhang et al., 2013; Lasley-Rasher et al., 2016; Steele et al., 2018a: Steele et al., 2018b), which has been the primary focus of previous toxicity studies with P. parvum. In the present study, behavioral alterations of fish were observed at cell densities below those inducing acute mortality and were one of the most sensitive endpoints examined.

The fish behavioral responses examined in the present study are utilized in ecotoxicology and biomedical applications, with a particular emphasis on zebrafish behavior for drug discovery applications. Recent advances in computational and tracking technologies allow for high throughput screening (HTS) of fish models, primarily during early life stages. Fish behavioral syndromes and phenotypes of a wide array of chemicals have been studied in order to associate specific behavioral alterations with chemical mode/mechanism of action (Drummond and Russom, 1990; Rihel et al., 2010). In the present study, swimming activity was significantly reduced in fathead minnow across the freezing and cruising speed thresholds indicating refractory responses after exposure to P. parvum grown under all experimental conditions. However, stimulatory behaviors were observed, suggesting that these decreases in behavior resulted from increased activity across the highest speed threshold. Chemical classes that also produced a hyperactivity syndrome in 30 d old fathead minnows include primary aliphatic amines, phenols and halogenated phenols (Drummond and Russom, 1990). These compounds elicit toxicity through disruption of metabolic activity and function, which is consistent with the proposed fish mode of action for P. parvum (Shilo, 1967). Our recent studies with the acetylcholinesterase inhibitor diazinon elicited similar stimulatory behavioral effects in zebrafish and fathead minnows (Steele et al. 2018b). Whether P. parvum affects neurotransmission is not known but deserves future attention.

5. Conclusions

P. parvum HABs can cause profound fish kills, and thus fish mortality responses in the laboratory and the field have increasingly been examined during ecology and toxicology studies. We report here a number of sensitive sublethal responses to P. parvum grown under conditions associated with HAB events in inland waters. To our knowledge this is the first study to examine OS gene expression and biochemical biomarkers and behavioral changes in fish following P. parvum exposure comparatively with two common fish models. Traditional biochemical markers indicated a contribution of OS to P. parvum toxicity. Induction of antioxidant gene expression confirmed these observations. Behavior alterations were also a sensitive indicator of P. parvum exposure. Future studies are necessary to identify causative toxin(s) associated with such sublethal responses and whether these molecule(s) are differentially produced across environmental gradients and other P. parvum strains.

Highlights:

• Though Prymnesium parvum causes fish kills, sublethal toxicity is not understood.

• We compared diverse responses of two common fish models to P. parvum exposure.

• Nonoptimally grown P. parvum elicited fish oxidative stress and behavioral effects.

• Both fish models exhibited differential sensitivities to P. parvum culture conditions.

• Assessment and management of P. parvum HABs should include sublethal responses.